Data Science & Machine Learning · Developer Education · Open Source

I'm a machine learning engineer and developer advocate — PhD in Computational Science, Pythonista since 2009. I like living at the intersection of two crafts: building data & ML systems, and teaching developers how they actually work.

• 🔬 Data Science & Machine Learning — deep learning for biomedicine and digital pathology, NLP, spatio-temporal forecasting. My favourite corner of ML is privacy-preserving machine learning: federated learning, differential privacy, and local / self-hosted AI
• 📣 Developer education — tutorials, workshops, docs, demos, talks. Explaining hard things clearly is a skill I practise as deliberately as writing code
• 🐍 Active in the Python community since 2009 — conferences, education, and a lot of open source
• 🧪 The organisations on my profile are where the fun happens: research labs, summer schools, and community projects I build with (MPBA, DynamicGenetics, webvalley, kubeflow-kale, and friends)
• 🎄 Every December I disappear into Advent of Code — Python, of course
• 🃏 Magic: The Gathering player — Premodern is my format (@lotus_vale on Discord). I'm the tech guy behind Chaos Orcs Fest and Bristol Premodern — both Django + PyScript (the deck-submission flow runs Python in the browser 🤯) — and I contribute to Forge, the open-source rules engine for MTG

Core & Scientific Python

ML & Deep Learning

LLMs & Local AI

Web, Data & Infra

Speaking and teaching are how I give back to the community — and honestly, how I learn best. I've been on stage at PyCon US / DE / IT, EuroPython, SciPy, EuroSciPy, PyData and more, talking about Python, machine learning, and privacy-preserving AI. All the slides live on Speaker Deck.

• 🧑‍🏫 Hands-on workshops on Python, deep learning & ML best practices — the messy, live-coding kind
• 🎓 Lecturer: University of Bristol, FBK Academy, WebValley Summer School · Carpentries certified instructor
• 🏅 Software Sustainability Institute Fellow (privacy-enhancing technologies for ML)

• private-llm-tailnet — chat with a self-hosted LLM from your own devices over a private Tailscale network: MLX model server on Apple silicon, single-file chat client, tailnet-only HTTPS
• mistral-concordance — cross-jurisdictional clinical guideline navigator built on Mistral Workflows with a hybrid local/cloud inference stack — pauses for clinician review on disagreement
• mlx-quant-bench — benchmarking LLM quantisation on Apple silicon with MLX
• transformers-laptop-bench — benchmark of open LLM inference cost on consumer hardware (CUDA / Apple silicon / CPU) with Hugging Face Transformers
• ml-regression-watch — benchmarking, numerical validation, and CI regression detection for ML models across execution configurations · 🆕 now with per-layer analysis on DistilBERT

Python & data science

• python-in-a-notebook — a whole collection of Jupyter notebooks on Python programming
• python-data-science — lecture notes & materials for a full Python data science course
• programming-for-data-science — programming for data science course (WebValley 2021)
• python-programming — Python programming @ WebValley 2019
• numpy-euroscipy — introduction to NumPy (EuroSciPy tutorial)

ML & deep learning

• deep-learning-keras-tensorflow — intro to deep neural networks with Keras & TensorFlow. One of the first Keras tutorials ever (EuroSciPy 2016) · ⭐ 3k
• pytorch-beautiful-ml-data — data patterns & OOP abstractions for PyTorch (PyData Global tutorial)
• deep-learning-health-life-sciences — workshop on deep learning for health & life sciences
• deep-unsupervised-learning — deep unsupervised learning course
• unsupervised-learning-tutorial — hands-on unsupervised learning tutorial
• ml-course — machine learning course taught at WebValley 2022

Privacy-preserving ML

• ppml-tutorial — hands-on privacy-preserving machine learning (SciPy, PyConDE, Mozilla Festival, EuroSciPy)
• privacy-preserving-data-science — full course on privacy-enhancing technologies & PPML (SSI Fellowship output)
• syft-heart-disease-tutorial — end-to-end federated learning sample app with PySyft/SyftBox

• The Learning Machine — interactive human-in-the-loop online machine learning that lets the public teach a machine to recognise human emotion — exhibited at the We The Curious science museum in Bristol
• notexbook-jupyter-theme — a Jupyter theme for LaTeX lovers and the typographically obsessed 🤓
• CovidResponseMap — interactive community-support mapping, adopted by Public Health Wales during the pandemic

• Chaos Orcs Fest & Bristol Premodern — tournament websites for the Premodern community, built with Django + PyScript for in-browser deck submission
• Forge — contributor to the open-source MTG rules engine
• deck-recognizer — PyScript-powered deck recogniser for Premodern tournaments
• mtg-collection-analysis — a data-science journey into my own MTG collection

Upstream contributions — documentation, tutorials, testing across the ML ecosystem: pytorch/pytorch · scikit-learn/scikit-learn · keras-team/keras · lmcinnes/umap (main docs/testing contributor) · pyscript/pyscript (since launch) · OpenMined/PySyft · Project-MONAI